2013 · 91 citations · 15 references
ComputingComputer ScienceComputational LinguisticsInformation RetrievalNatural Language ProcessingKeyword ExtractionLanguage ModelingMachine TranslationPattern RecognitionWord RecognitionKeyword SpottingNormalizationScore Normalization
We present two techniques that are shown to yield improved Keyword Spotting (KWS) performance when using the ATWV/MTWV performance measures: (i) score normalization, where the scores of different keywords become commensurate with each other and they more closely correspond to the probability of being correct than raw posteriors; and (ii) system combination, where the detections of multiple systems are merged together, and their scores are interpolated with weights which are optimized using MTWV as the maximization criterion. Both score normalization and system combination approaches show that significant gains in ATWV/MTWV can be obtained, sometimes on the order of 8-10 points (absolute), in five different languages. A variant of these methods resulted in the highest performance for the official surprise language evaluation for the IARPA-funded Babel project in April 2013.
15
<i>Numerical Recipes, The Art of Scientific Computing</i>
William H. Press, Brian P. Flannery, Saul A. Teukolsky et al. · American Journal of Physics · 1987 · 10.1K citations
Sequence-discriminative training of deep neural networks
Karel Veselý, Arnab Ghoshal, Lukáš Burget et al. · 2013 · 662 citations · Full text
Numerical recipes: The art of scientific computing
Peter Frolkovič · Acta Applicandae Mathematicae · 1990 · 566 citations
Rapid and accurate spoken term detection
David R. Miller, Michael Kleber, Chia-Lin Kao et al. · 2007 · 254 citations